DALL-E 3 is live through Bing Image Creator


OIG.xAygoaZNXBSfuFEs5dSc

OIG.CmY9yCp8FxEA6b7ZhEnA

OIG.YqkeopNKmBE5y_J_ww.D
 
Any suggestions for image upscaling? It's either messing with the details and smothering them out or it's perfect and highly priced.

 
Thank you very much for this! Worked wonderful and was easy to setup. You made my day!
 
I tried to recreate the paperback cover for A Spell for Chameleon.
Dall-E's idea of a Manticore doesn't really agree with Michael Whelan's.

OUSzB7R.jpg


Original:
EnRr7DR.jpg
 
So I have been able to defeat the censorship on occasion and get some pretty cool stuff. Still cant believe how prudish it is. Anyways the 1024x1024 image it puts out is unfortunately pretty low res for the prompts I am attempting so I lose some detail through upscaling and fixing some of the bad parts of images.

As an example here is the raw output on a prompt I put in (Basically Dark Fantasy, Epic Oil on Canvas with stuff I've learned from SD added in)

QrpIpf3.jpg


I then used subtle diffusion 1.5 to fix her face and give it more detail. After that I upscaled the image and did some tweaks in gimp.
UYZ6mbC.png


Anyways, would love to have Dall-e 3 generation with out the prude filter and with the flexibility of SD prompts + negatives.


Here are a couple more raw images to show the stuff I am getting



CEYRiZ1.jpg
cltn7IS.jpg
gcCK3pM.jpg
mV2aCZc.jpg

kHZULmw.jpg
TYbqaaT.jpg

6Q7UiQU.jpg
7unPmyU.jpg



I have a metric crap tone more that I will share later
 
Last edited:
So I have been able to defeat the censorship on occasion and get some pretty cool stuff. Still cant believe how prudish it is. Anyways the 1024x1024 image it puts out is unfortunately pretty low res for the prompts I am attempting so I lose some detail through upscaling and fixing some of the bad parts of images.

As an example here is the raw output on a prompt I put in (Basically Dark Fantasy, Epic Oil on Canvas with stuff I've learned from SD added in)

QrpIpf3.jpg


I then used subtle diffusion 1.5 to fix her face and give it more detail. After that I upscaled the image and did some tweaks in gimp.
UYZ6mbC.png


Anyways, would love to have Dall-e 3 generation with out the prude filter and with the flexibility of SD prompts + negatives.


Here are a couple more raw images to show the stuff I am getting



CEYRiZ1.jpg
cltn7IS.jpg
gcCK3pM.jpg
mV2aCZc.jpg

kHZULmw.jpg
TYbqaaT.jpg

6Q7UiQU.jpg
7unPmyU.jpg



I have a metric crap tone more that I will share later
I've noticed that even without trying, it'll just sometimes give me ass cheek. Which is so weird considering how prudish it is.
 
Last edited:
I've noticed that even without trying, it'll just sometimes give me ass cheek. Which is so weird considering how prudish it is.
You know what's hilarious, I also have access to Dall-e 3 through the chatgpt app and it is even more censored.

This prompt right here got flagged: "Digital Art: modern Caucasian woman wearing black leggings and a light blue crop top, tactical belts, and a sword, fighting an eldritch horror with magic, ultra realistic" and so I asked it to rewrite the prompt to follow the content policy and all four of the prompts it posted also got flagged...I mean this shit is just a meme at this point. Its also hilarious to watch ChatGPT woke-ify my prompts right in front of me. Sexy, pretty, or any descriptive that denotes attractiveness gets replaced with "Strong, confident, resolute, sturdy, etc"... Like what a fucking stereotype.

Anyways here is some more eye food. Enjoy while it lasts because the ai is learning and my ability to get around big brother is slowly fading away.

aqdkcjO.jpg
iG8x3Z5.jpg
whi3zGV.jpg
XQdgr0s.jpg
3hTYGiM.jpg
naprwkp.jpg
w6fBeeP.jpg
yytpjmV.jpg
nOnEusY.jpg
6eQ3RcX.jpg
INzKo3X.jpg
Euh24Ix.jpg
vzPtj0C.jpg
 
Last edited:
Everything is blocked or "Unsafe", Nothing I am asking is sexual or Gorey or have any real life people in it too
Borderline unusable

My work around for getting a fat guy is to just say "john candy" right now, it doesn't seem to know anything about John Candy except that he was a fat white guy, which is not quite fat enough for generating a Batman as fat as I'd like, which I desperately need for a project I'm working on.

qUIxsk2.png
 
I hope the open source side of things gives rise to a free for all in terms of creativity and subject matter, to an extent. The great thing artists have is their minds eye and combinations of creative sparks and/or process to produce their art.

I'm just as interested in various levels and types of artists' creations as I am with what everyday people or elderly/kids/cultures create.
 
Last edited:
OIG.ovvHtqIMfEMjMgMfkNFF

OIG.xIWXf0iChUhpBgJX_csN

The main protagonist in my first finished short story. Still working on the second story, it's gone longer than I thought it would but DALL-E got her down too when I was allowed to use the word Fat
OIG.ijVGNt7msHHGY2ZxEwPb
 
Last edited:
This is a scene in a Jesus and Buddha anime.

"Jesus and Buddha relaxing at a sauna"

tYiiHoF.jpg
luiso5P.jpg
MpFV2sl.jpg


Not from an anime.

"Jesus and skeletor wearing spacesuits riding a rollercoaster on the moon"

I only added "spacesuits" because my first attempt was blocked.

D4BcdmD.jpg
IX0SWjR.jpg
e79Gx5Q.jpg
ME0LsFd.jpg
 
Last edited:
It's getting fucking ridiculous to create anything without some unsafe image content warning. It's lost its appeal for me. I think I'm done with it for a while.
 
Top Bottom